[Feature] Benchmark loading of production deployments

[vicsn]

🚀 Feature

We should move this test into snarkVM and run it as a benchmark: https://github.com/ProvableHQ/aleo-program-regressions/tree/master/regressions/profile_process_load_from_store

If possible, it would be nice to run 3 separate benchmarks, one for canary, testnet and mainnet.

Downloading the programs and deployments should not be part of the benchmark, ideally the deployments can be cached ahead of time in a GCP instance / image.

Note that over time, more programs are added to these network. This has some implications:

• We may want to manually periodically update the GCP instance / image to contain all of the latest programs
• We may want to alert not on total runtime but on total runtime divided by the total number of programs

@kaimast since you have experience with the recent snarkOS benches, you'll probably be best equipped to tackle this one at the moment. :)